Organic light-emitting diodes (OLEDs) constitute a new class of emissive devices, which present high efficiency and low voltage operation, among other advantages over current technology. Multilayer architecture (M-OLED) is generally used to optimize these devices, specially overcoming the suppression of light emission due to the exciton recombination near the metal layers. However, improvement in recombination, transport and charge injection can also be achieved by blending electron and hole transporting layers into the same one. Graded emissive region devices can provide promising results regarding quantum and power efficiency and brightness, as well. The massive number of possible model configurations, however, suggests that a search algorithm would be more suitable for this matter. In this work, multilayer OLEDs were simulated and fabricated using Genetic Algorithms (GAs) as evolutionary strategy to improve their efficiency. Genetic Algorithms are stochastic algorithms based on genetic inheritance and Darwinian strife to survival. In our simulations, it was assumed a 50 nm width graded region, divided into five equally sized layers. The relative concentrations of the materials within each layer were optimized to obtain the lower V/J(0.5) ratio, where Vis the applied voltage and J the current density. The best M-OLED architecture obtained by genetic algorithm presented a V/J(0.5) ratio nearly 7% lower than the value reported in the literature. In order to check the experimental validity of the improved results obtained in the simulations. two M-OLEDs with different architectures were fabricated by thermal deposition in high vacuum environment. The results of the comparison between simulation and some experiments are presented and discussed. (C) 2009 Elsevier B.V. All rights reserved.